Transit Priority

import pandas as pd
import matplotlib.pyplot as plt
import geopandas as gpd
import shapely
import folium
from folium import Choropleth

Adding Vulnerability to Bus Metrics Analysis

To evaluate the relationship between transit service provision and community needs, the Transit Vulnerability Index (TVI) was spatially joined with bus network data aggregated by census tracts. This integration enabled a deeper analysis of how key transit metrics—such as frequency, headway, and ridership—align with areas of heightened social vulnerability.

#Import Data:
peak_metrics_network = gpd.read_file("Data/peak_metrics_network.geojson")
vulnerability = gpd.read_file("Data/vulnerability.geojson")
bus_network_philadelphia = gpd.read_file("Data/bus_network_philadelphia.geojson")

peak_metrics_network = peak_metrics_network.to_crs("EPSG: 2272")
vulnerability = vulnerability.to_crs("EPSG: 2272")

# Rename columns in the GeoDataFrames to avoid conflicts
if 'index_left' in peak_metrics_network.columns or 'index_right' in peak_metrics_network.columns:
    peak_metrics_network = peak_metrics_network.rename(columns={'index_left': 'pm_index_left', 'index_right': 'pm_index_right'})

if 'index_left' in vulnerability.columns or 'index_right' in vulnerability.columns:
    vulnerability = vulnerability.rename(columns={'index_left': 'vul_index_left', 'index_right': 'vul_index_right'})

# Perform the spatial join
final_segments = gpd.sjoin(
    peak_metrics_network,  
    vulnerability,         
    how="left",            
    predicate="intersects")

# Convert back to CRS
final_segments = final_segments.to_crs("EPSG: 4326")
final_segments = final_segments.sort_values(by='Social_Vulnerability_Index', ascending=False)
final_segments.head()

	pm_index_right	stop_id	bus_arrivals	frequency	headway	stop_name	stop_lat	stop_lon	location_type	parent_station	...	name10	n_veryhigh	year	hsi_score	hei_score	hvi_score	objectid	Shape__Area	Shape__Length	Social_Vulnerability_Index
16654	4767.0	15973.0	46.0	15.333333	3.913043	Broad St & Wyoming Av	40.024676	-75.147927	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
16443	4806.0	16894.0	25.0	8.333333	7.200000	Germantown Av & Windrim Av	40.022719	-75.158881	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
27967	4828.0	31460.0	25.0	8.333333	7.200000	Germantown Av & Berkley St	40.023818	-75.159409	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
16407	4823.0	16892.0	25.0	8.333333	7.200000	Germantown Av & Dennie St	40.020448	-75.156774	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
27968	4828.0	31460.0	25.0	8.333333	7.200000	Germantown Av & Berkley St	40.023818	-75.159409	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0

5 rows × 49 columns

Vulnerability on the bus network

We have generated a map to visualize transit rider vulnerability across the bus network, integrating the Social Vulnerability Index (SVI) to highlight areas of concern. The analysis reveals that West Philadelphia and North Philadelphia exhibit the highest levels of social vulnerability.

fig, ax = plt.subplots(figsize=(10, 8))
final_segments.plot(
    ax=ax,
    column="Social_Vulnerability_Index",  
    cmap="plasma",        
    legend=True,           
    markersize=1         
)

plt.title("Vulnerability by Street Segment", fontsize=16)
plt.xlabel("Longitude")
plt.ylabel("Latitude")

Text(118.3577266549966, 0.5, 'Latitude')

Criteria for Critical Bus Network Segments Improvements

To prioritize bus transit solutions effectively, we established the following criteria for determining high-priority areas:

Transit Vulnerability Index: Areas must have a Transit Vulnerability Index (TVI) score of 4 or higher, highlighting significant levels of social vulnerability.

Headway: Routes with headways under 2 minutes are considered, ensuring focus on high-frequency services where improvements can maximize impact.

Ridership: Only routes within the upper quartile of ridership are included, reflecting areas of high demand and usage.

These criteria allow us to target interventions where they are most needed, addressing equity, efficiency, and demand in a data-driven manner.

upper_quartile = final_segments["Ridership"].quantile(0.25)

filtered_segments = final_segments[
    (final_segments["Social_Vulnerability_Index"] >= 3) &  
    (final_segments["headway"] < 5) &                    
    (final_segments["Ridership"] >= upper_quartile)]

filtered_segments.head()

	pm_index_right	stop_id	bus_arrivals	frequency	headway	stop_name	stop_lat	stop_lon	location_type	parent_station	...	name10	n_veryhigh	year	hsi_score	hei_score	hvi_score	objectid	Shape__Area	Shape__Length	Social_Vulnerability_Index
16654	4767.0	15973.0	46.0	15.333333	3.913043	Broad St & Wyoming Av	40.024676	-75.147927	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
16389	4767.0	15973.0	46.0	15.333333	3.913043	Broad St & Wyoming Av	40.024676	-75.147927	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
16388	4767.0	15973.0	46.0	15.333333	3.913043	Broad St & Wyoming Av	40.024676	-75.147927	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
16386	4767.0	15973.0	46.0	15.333333	3.913043	Broad St & Wyoming Av	40.024676	-75.147927	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0
16653	4767.0	15973.0	46.0	15.333333	3.913043	Broad St & Wyoming Av	40.024676	-75.147927	NaN	NaN	...	280.0	1.0	2023.0	6.741606	0.974944	4.672679	344.0	1.226439e+06	4682.828235	6.0

5 rows × 49 columns

Visualising the most important segments for improvements:

The segments selected based on the criteria outlined above are concentrated along major streets in Philadelphia, including Broad Street, Market Street, Kensington Avenue, and Roosevelt Boulevard. High ridership on these segments likely reflects passenger transfers from nearby subway stations.

from shapely.geometry import Point

# Re-project to a projected CRS for accurate centroid calculation
projected_segments = filtered_segments.to_crs(epsg=3857)

# Calculate the mean centroid coordinates in the projected CRS
mean_y = projected_segments.geometry.centroid.y.mean()
mean_x = projected_segments.geometry.centroid.x.mean()

# Convert the mean centroid coordinates back to WGS84 for Folium
mean_point = gpd.GeoSeries([Point(mean_x, mean_y)], crs=3857).to_crs(epsg=4326)
mean_lat, mean_lon = mean_point.geometry[0].y, mean_point.geometry[0].x

# Initialize a Folium map centered around the data
m = folium.Map(
    location=[mean_lat, mean_lon],
    zoom_start=11,
    tiles='CartoDB Positron')

# choropleth map for TVI
folium.Choropleth(
    geo_data=vulnerability,
    name="Social Vulnerability Score",
    data=vulnerability,
    columns=["GEOID", "Social_Vulnerability_Index"],
    key_on="feature.properties.GEOID", 
    fill_color="YlOrRd",
    fill_opacity=0.3,
    line_opacity=0.1,
    legend_name="Social Vulnerability Score",
).add_to(m)

folium.GeoJson(
    bus_network_philadelphia,
    style_function=lambda x: {"color": "grey", "weight": 0.5, "opacity": 0.4},
    name="Bus Network (Philadelphia)"
).add_to(m)

folium.GeoJson(
    filtered_segments,
    style_function=lambda x: {"color": "blue", "weight": 3, "opacity": 1},
    name="Filtered Segments"
).add_to(m)

folium.LayerControl().add_to(m)

m

Make this Notebook Trusted to load map: File -> Trust Notebook

The table below showcases the top 10 bus segments ranked by ridership. These segments represent the busiest parts of the network, where high passenger volumes underscore their critical role in Philadelphia’s transit system. Prioritizing improvements in these areas is essential to enhance service reliability and meet the needs of the city’s most active transit users. By focusing on these high-demand corridors, transit agencies can address existing inefficiencies and better align services with the needs of vulnerable communities.

# Sort the GeoDataFrame by the 'headway' column in ascending order
filtered_segments = filtered_segments.sort_values(by='headway', ascending=True)

# Group by 'stop_name' and calculate aggregate values for the selected columns
final_streets = (
    filtered_segments.groupby("stop_name")[["headway", "frequency", "Ridership", "Social_Vulnerability_Index"]]
    .mean() 
    .reset_index())

final_streets.rename(columns={"Social_Vulnerability_Index": "Transit_Rider_Vulnerability_Index"}, inplace=True)
final_streets = final_streets.sort_values(by='Ridership', ascending=False)
final_streets.head(10)

	stop_name	headway	frequency	Ridership	Transit_Rider_Vulnerability_Index
10	Arrott Transportation Center	2.903226	20.666667	1550.0	3.000000
1	23rd St & Venango St Loop	1.855670	32.333333	1503.0	4.000000
26	Broad St & Olney Av - FS	3.750000	16.000000	825.0	3.750000
63	Roosevelt Blvd & Broad St - FS	3.913043	15.333333	666.0	3.000000
30	Broad St & Tabor Rd	3.461538	17.333333	665.0	3.285714
47	Market St & 40th St	4.090909	14.666667	575.0	3.000000
2	33rd St & Dauphin St Loop	4.864865	12.333333	536.0	4.000000
35	Frankford Av & Margaret St	3.529412	17.000000	492.0	3.000000
73	Woodland Av & 50th St	3.600000	16.666667	481.0	5.000000
18	Broad St & Allegheny Av	3.471429	17.333333	415.0	4.000000